In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
import plotly.express as px
In [2]:
df = pd.read_csv("LS_2.0 (1).csv")
In [3]:
pd.set_option('display.max_rows',20000, 'display.max_columns',100)
In [4]:
df.shape
Out[4]:
In [5]:
df.info()
In [6]:
df.head(5)
Out[6]:
In [7]:
df.isnull().sum().sort_values(ascending=False).head(30)
Out[7]:
In [8]:
df[df['NAME']=='NOTA'].head()
Out[8]:
In [9]:
df = df.fillna({'SYMBOL':'NO SYMBOL',
'GENDER':'NOT APPLICABLE',
'CRIMINAL\nCASES':0,'AGE':0.0,
'CATEGORY':'NOT APPLICABLE',
'EDUCATION':'NOT APPLICABLE',
'ASSETS':'Rs 0',
'LIABILITIES':'Rs 0',})
In [10]:
df.isnull().sum().sort_values(ascending=False).head(30)
Out[10]:
In [11]:
(df['PARTY'].value_counts()/len(df['PARTY']))*100
Out[11]:
In [12]:
df.head()
Out[12]:
In [13]:
(df['WINNER'].value_counts()/len(df['WINNER']))*100
Out[13]:
In [14]:
plt.rcParams['figure.figsize'] = (12,8)
labels = 'LOST', 'WON'
sizes = [76,24]
explode = (0, 0.1)
fig1, ax1 = plt.subplots()
ax1.pie(sizes, explode=explode, labels=labels, autopct='%0.0f%%',
shadow=True, startangle=90,center=(0, 0))
ax1.axis('equal') # Equal aspect ratio ensures that pie is drawn as a circle.
plt.title('2019 GENERAL ELECTION WINNING-LOSING PERCENTAGE')
plt.show()
In [15]:
plt.figure(figsize=(10,8))
sns.countplot(y='EDUCATION',hue='WINNER',data=df,palette="GnBu")
Out[15]:
In [16]:
plt.figure(figsize=(10,8))
sns.countplot(x='WINNER',hue='GENDER',data=df)
Out[16]:
In [17]:
#Filling Nan Values in CRIMINAL CASE COLUMN
df['CRIMINAL\nCASES']=df['CRIMINAL\nCASES'].replace(to_replace ="Not Available",
value ="0")
df['CRIMINAL\nCASES'] = df['CRIMINAL\nCASES'].astype(int)
In [18]:
df.head()
Out[18]:
In [19]:
#Checking if the candidate is criminal/ncase or not
for i in range(len(df)):
if df.iloc[i,7]>0:
df.iloc[i,7]='HAS CASE'
else:
df.iloc[i,7]='NO CASE'
In [20]:
plt.figure(figsize=(8,4))
sns.countplot(x='WINNER',hue='CRIMINAL\nCASES',data=df)
Out[20]:
In [21]:
sns.countplot(y='GENDER',hue='CRIMINAL\nCASES',data=df,palette='hot')
Out[21]:
In [22]:
df.head()
Out[22]:
In [23]:
df[df['ASSETS']=='Not Available']
Out[23]:
In [24]:
for i in range(len(df)):
if df.iloc[i,11]=='Not Available':
df.iloc[i,11] = "-1"
In [25]:
#Removing '\n ~' from ASSETS
df['ASSETS'] = df['ASSETS'].str.split('\n ~', 1, expand=True)[0]
#Removing 'Rs' from ASSETS
df['ASSETS'] = df['ASSETS'].str.split(expand=True)[1]
#Removing special charater "," and then joinig it
df['ASSETS']=df['ASSETS'].str.split(",").str.join(" ")
#Removing the space between the join
df['ASSETS']=df['ASSETS'].str.replace(' ', '')
#Converting ASSETS column into flaot as int values are very high
df['ASSETS'] = df['ASSETS'].astype(float)
In [26]:
df['ASSETS'] = df['ASSETS'].fillna(value = 0.0)
In [27]:
STATUS = []
for i in df['ASSETS']:
if i >0.0 and i < 500000.0:
STATUS.append('NEAR TO BPL')
if i >= 500000.0 and i <= 1000000.0:
STATUS.append('LOWER CLASS')
if i >= 1000001.0 and i <= 2500000.0:
STATUS.append('LOWER MIDDLE CLASS')
elif i >= 2500001.0 and i <= 10000000.0:
STATUS.append('MIDDLE CLASS')
elif i >= 10000001.0 and i <= 100000000.0:
STATUS.append('UPPER MIDDLE CLASS')
elif i >= 100000000.0 and i <= 250000000.0:
STATUS.append('ELITE CLASS')
elif i >= 250000001.0 and i <= 1000000000.0:
STATUS.append('SUPER RICH')
elif i >= 1000000001.0:
STATUS.append('RICHEST OF RICH')
elif i == -1.0:
STATUS.append('ASSETS NOT MENTIONED')
elif i == 0.0:
STATUS.append('NO ASSETS')
df['STATUS'] = STATUS
In [28]:
sns.countplot(y='STATUS',hue='CRIMINAL\nCASES',data=df,palette='hot')
Out[28]:
In [29]:
print(df['PARTY'].nunique(),df['PARTY'].unique())
In [30]:
#Using Groupby on Party with Total Votes(TO CHECK HIGHEST PERCENTAGE OF VOTES FOR TOP 10 PARTIES)
p = df.groupby('PARTY')['TOTAL\nVOTES']
X = ((p.sum()/df['TOTAL\nVOTES'].sum())*100).sort_values(ascending = False).head(10)
X
Out[30]:
In [31]:
plt.rcParams['figure.figsize'] = (12,8)
labels = 'BJP', 'INC','AITC','BSP','SP','YSRCP','CPI(M)','DMK','SHS','TDP','OTHERS'
sizes = [38,20,4,3,2,2,2,2,2,2,23]
colors=('orange', 'green', 'deeppink', 'blue', 'red', 'yellow', 'crimson', 'brown','darkorange','pink','gray')
my_circle = plt.Circle((0, 0), 0.7, color='white')
d = plt.pie(sizes, labels=labels, autopct='%0.0f%%',
startangle=90,colors=colors, labeldistance=1.05)
plt.axis('equal')
plt.gca().add_artist(my_circle)
plt.title('VOTE SHARE OF PARTY')
plt.show()
In [32]:
df.head()
Out[32]:
In [33]:
df['AGE'].nunique()
Out[33]:
In [34]:
df.head()
Out[34]:
In [35]:
plt.rcParams['figure.figsize'] = (10,8)
labels = 'BJP','INC','DMK','YSRCP','AITC','SHS','JD(U)','BSP','BJD','TRS','OTHER'
sizes = [300, 52, 23, 22, 22, 18, 16, 11, 11, 9,55]
colors=('orange', 'green', 'deeppink', 'blue', 'red', 'yellow', 'crimson', 'brown','darkorange','pink','gray')
my_circle = plt.Circle((0, 0), 0.7, color='white')
d = plt.pie(sizes, labels=labels, autopct='%0.0f%%',
startangle=90,colors=colors, labeldistance=1.05)
plt.axis('equal')
plt.gca().add_artist(my_circle)
plt.title('OUT OF 539 SEATS WINNER DISTRIBUTION')
plt.show()
In [36]:
df.head()
Out[36]:
In [37]:
p = df.groupby('STATE')['CONSTITUENCY'].nunique().reset_index()
plt.figure(figsize=(15,10))
sns.barplot(y='STATE',x='CONSTITUENCY',data=p,palette='spring')
Out[37]:
In [38]:
p = df.groupby(['PARTY','EDUCATION','GENDER'])['WINNER'].sum().reset_index().sort_values('WINNER',ascending = False)
p = p[p['WINNER']!=0]
fig = px.bar(p, x='EDUCATION', y='WINNER',hover_data =['PARTY'], color='GENDER', height=650)
fig.show()
In [39]:
p = df.groupby(['PARTY','STATE']).sum().reset_index().sort_values('PARTY',ascending =True)
fig = px.bar(p, x='STATE', y='WINNER',hover_data =['PARTY'], color='PARTY', height=650)
fig.show()
In [40]:
#Creating Age_Group from Age column
AGE_GROUP = []
for i in df['AGE']:
if i >= 24 and i <=35:
AGE_GROUP.append('YOUNG AGE')
elif i >= 36 and i<=60:
AGE_GROUP.append('MIDDLE AGE')
elif i >=60:
AGE_GROUP.append('OLD AGE')
else:
AGE_GROUP.append('NOT KNOWN')
df['AGE_GROUP'] = AGE_GROUP
In [41]:
p = df.groupby(['PARTY','AGE_GROUP'])['WINNER'].sum().reset_index().sort_values('WINNER',ascending = False)
p = p[p['WINNER']!=0]
fig = px.bar(p, x='PARTY',y='WINNER',hover_data =['AGE_GROUP'], color='AGE_GROUP', height=650)
fig.show()
In [42]:
p = df.groupby(['PARTY','GENDER'])['WINNER'].count().reset_index().sort_values('WINNER',ascending = False)
#p = p[p['WINNER']!=0]
fig = px.bar(p, x='PARTY',y='WINNER',hover_data =['GENDER'], color='GENDER', height=700)
fig.show()
In [43]:
p = df.groupby(['PARTY','CRIMINAL\nCASES'])['WINNER'].sum().reset_index().sort_values('WINNER',ascending = False)
p = p[p['WINNER']!=0]
fig = px.bar(p, x='PARTY',y='WINNER',hover_data =['CRIMINAL\nCASES'], color='CRIMINAL\nCASES', height=750)
fig.show()
In [44]:
from sklearn.preprocessing import LabelEncoder
labelEncoder_X = LabelEncoder()
df['STATE'] = labelEncoder_X.fit_transform(df['STATE'])
df['CONSTITUENCY'] = labelEncoder_X.fit_transform(df['CONSTITUENCY'])
df['NAME'] = labelEncoder_X.fit_transform(df['NAME'])
df['PARTY'] = labelEncoder_X.fit_transform(df['PARTY'])
df['SYMBOL'] = labelEncoder_X.fit_transform(df['SYMBOL'])
df['GENDER'] = labelEncoder_X.fit_transform(df['GENDER'])
df['CRIMINAL\nCASES'] = labelEncoder_X.fit_transform(df['CRIMINAL\nCASES'])
df['CATEGORY'] = labelEncoder_X.fit_transform(df['CATEGORY'])
df['EDUCATION'] = labelEncoder_X.fit_transform(df['EDUCATION'])
df['STATUS'] = labelEncoder_X.fit_transform(df['STATUS'])
df['AGE_GROUP'] = labelEncoder_X.fit_transform(df['AGE_GROUP'])
In [45]:
X=df.drop(['WINNER','ASSETS','LIABILITIES','GENERAL\nVOTES','POSTAL\nVOTES','AGE','OVER TOTAL ELECTORS \nIN CONSTITUENCY','OVER TOTAL VOTES POLLED \nIN CONSTITUENCY'],axis=1)
y=df['WINNER']
In [46]:
from sklearn.model_selection import train_test_split
X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.3,random_state=42)
In [47]:
from sklearn.preprocessing import StandardScaler
Scaler_X = StandardScaler()
X_train = Scaler_X.fit_transform(X_train)
X_test = Scaler_X.transform(X_test)
In [48]:
from sklearn.metrics import confusion_matrix, accuracy_score
In [49]:
#Logistic Regression
from sklearn.linear_model import LogisticRegression
lr = LogisticRegression()
lr.fit(X_train,y_train)
y_pred = lr.predict(X_test)
print(accuracy_score(y_test,y_pred))
print(confusion_matrix(y_test,y_pred))
In [50]:
#Decision Tree
from sklearn.tree import DecisionTreeClassifier
dtree = DecisionTreeClassifier()
dtree.fit(X_train,y_train)
predictions = dtree.predict(X_test)
print(accuracy_score(y_test,predictions ))
print(confusion_matrix(y_test,predictions ))
In [51]:
# Random Forest
from sklearn.ensemble import RandomForestClassifier
rfc = RandomForestClassifier(n_estimators=200)
rfc.fit(X_train,y_train)
rfc__pred = rfc.predict(X_test)
print(accuracy_score(y_test,rfc__pred))
print(confusion_matrix(y_test,rfc__pred))
In [52]:
import keras
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation
from keras.optimizers import SGD
In [53]:
from keras.wrappers.scikit_learn import KerasClassifier
# Doesn't work?
# class TwoLayerFeedForward:
# def __call__():
# clf = Sequential()
# clf.add(Dense(9, activation='relu', input_dim=3))
# clf.add(Dense(9, activation='relu'))
# clf.add(Dense(3, activation='softmax'))
# clf.compile(loss='categorical_crossentropy', optimizer=SGD())
# return clf
def twoLayerFeedForward():
clf = Sequential()
clf.add(Dense(9, activation='relu', input_dim=3))
clf.add(Dense(9, activation='relu'))
clf.add(Dense(3, activation='softmax'))
clf.compile(loss='categorical_crossentropy', optimizer=SGD(), metrics=["accuracy"])
return clf
# clf = KerasClassifier(TwoLayerFeedForward(), epochs=100, batch_size=500, verbose=0)
clf = KerasClassifier(twoLayerFeedForward, epochs=100, batch_size=500, verbose=0)
In [54]:
from sklearn.ensemble import AdaBoostClassifier
from sklearn.datasets import make_classification
In [66]:
clf = AdaBoostClassifier(n_estimators=100, random_state=0)
clf.fit(X_train, y_train)
AdaBoostClassifier(n_estimators=100, random_state=0)
clf.predict(X_test)
clf.score(X_test, y_test)
Out[66]:
In [62]:
from sklearn.ensemble import GradientBoostingClassifier
In [57]:
clf = GradientBoostingClassifier(n_estimators=100, learning_rate=1.0,max_depth=1, random_state=0).fit(X_train, y_train)
clf.score(X_test, y_test)
Out[57]:
In [58]:
from keras.models import Sequential
from keras.layers import Dense
from keras.wrappers.scikit_learn import KerasClassifier
from sklearn.model_selection import cross_val_score
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import StratifiedKFold
In [60]:
encoder = LabelEncoder()
encoder.fit(y_train)
encoded_Y = encoder.transform(y_train)
# baseline model
def create_baseline():
# create model
model = Sequential()
model.add(Dense(13, input_dim=13, activation='relu'))
model.add(Dense(1, activation='sigmoid'))
# Compile model
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
return model
# evaluate model with standardized dataset
estimator = KerasClassifier(build_fn=create_baseline, epochs=100, batch_size=5, verbose=0)
kfold = StratifiedKFold(n_splits=10, shuffle=True)
results = cross_val_score(estimator, X_train, encoded_Y, cv=kfold)
print("Baseline: %.2f%% (%.2f%%)" % (results.mean()*100, results.std()*100))
In [ ]: